Compositional Generative Modeling: A Single Model is Not All You Need
This work addresses the challenge of building scalable and adaptable generative AI systems for researchers and practitioners, offering a novel paradigm shift from monolithic models.
The paper tackles the problem of data inefficiency and lack of generalization in large monolithic generative models by proposing a compositional approach that composes smaller generative models, enabling more data-efficient learning and generalization to unseen data distributions.
Large monolithic generative models trained on massive amounts of data have become an increasingly dominant approach in AI research. In this paper, we argue that we should instead construct large generative systems by composing smaller generative models together. We show how such a compositional generative approach enables us to learn distributions in a more data-efficient manner, enabling generalization to parts of the data distribution unseen at training time. We further show how this enables us to program and construct new generative models for tasks completely unseen at training. Finally, we show that in many cases, we can discover separate compositional components from data.